My 274-game AI-only Civilization V experiment: The definitive ranking

I spent about 1000 hours running 274 AI-only games in Civ5. Here’s how I did it:

  • 22 random civs on a huge shuffle map
  • AI difficulty, no start bias, random personalities
  • Time victory only, 500 turn limit
  • Ranked civs based on final standings (22 points for 1st, 1 for last)

Results were surprising! Ethiopia came out on top as nearly unbeatable. Their early religion and defensive bonuses make them tough to crack. The Iroquois were unexpectedly strong too - they just kept pumping out cities.

On the flip side, poor Mongolia ended up dead last. Their aggression backfired badly. Japan and Sweden also struggled, with bonuses that didn’t seem to help much.

What do you think of these results? Any shockers or outcomes you expected? I’d love to hear your thoughts on the strongest and weakest civs!

That’s some serious dedication! Ethiopia as the top dog is wild. I never really gave them much thought when playing. Kinda funny how the AI handles different civs compared to humans. Makes me wanna go back and try Ethiopia now.

Man, that’s an epic experiment. I’m blown away by Ethiopia crushing it. Never would’ve guessed they’d be top dog. Crazy how the AI handles different civs compared to humans. Mongolia tanking makes sense though. The AI probably just Yolo’d their units without any real strategy. Really cool data, makes me wanna fire up Civ5 again and try some new civs.

I’m amazed at the dedication! 1000 hours is no joke.

Ethiopia’s dominance is surprising. Never thought their bonuses were that strong.

Poor Mongolia though. AI probably doesn’t know how to use those horses properly.

Wow, that’s a crazy amount of time to put into this! I’m surprised Ethiopia came out on top. I always thought Babylon or Korea would dominate with their science boosts. Kinda makes sense about Mongolia though. The AI probably doesn’t use their special units well. Cool experiment!

That’s a ton of data! Interesting to see Ethiopia dominate.

I wonder how different map types might shake up the rankings.

Have you thought about trying this with different victory conditions?